Optimization of Blastability Parameters of Ore Rock Based on Multi-source Information Fusion Theory

In this paper, the Rough Set (RS) theory is adopted to reduce the rock blasting parameters and the RS-BPNN and the RS-SVM rock blastability prediction models are established respectively. The factors, such as the volume of the blasting crater, the density of the ore rock, the elastic wave impedance of the ore rock, the boulder yield, the small yield, the average pass rate and the elastic wave velocity of the ore are obtained. Those 56 sets of data are normalized. The six attributes are reduced by the Rough Set theory, indicating that the average pass rate is a redundant factor, and it was removed. In terms of rock blastability index prediction, the average relative error of BPNN, RS-BPNN, SVM and RS-SVM is 9.68%, 7.29%, 1.84% and 1.71%. The study results show that the average pass rate is a redundant factor and the reduced model has a more obvious improvement on the prediction accuracy.